Introduction to AzureML

Level 300

(1)

Overview

This course explores Microsoft Azure’s AzureML service offering for students that are either new to machine learning or new to Azure. The course starts with an introduction to various aspects of building and “experiments” in AzureML and using MLStudio to create cohesive machine learning workflows. Each topic looks at different aspects of AzureML as well as introduces different concepts in machine learning such as regression, clustering and classification and when you would use each. The course moves onto more advanced topics such as how the R language can be used to enrich AzureML as well as being able to define neural networks and lastly how to integrate into more complex data orchestrations involving other services in Azure.

Course Audience

  • Data Scientist/Developer

Module 1: Introduction to AzureML and MLStudio

In this module you will learn how to create an AzureML workspace and create new experiments. You will learn how to share experiments with your colleagues and customize your workspace. You will learn how to use MLStudio and save changes in your experiments.

In this lab, you will create a Preview AzureML Account and navigate through ML Studio.

In this lab you will create a blank experiment and run a sample experiment.

Module 2: Reading and Writing Data

In this module you will learn how to read and write data to and from a database, storage and Azure tables. You will be able to read data from the web through HTTP and OData and upload datasets from your local machine or consume them from Azure storage.

In this lab you will read data from an OData and Web feed

In this lab you will create an Azure Table, insert some data and then read that data back from AzureML

In this lab you will version files within an experiment

In this lab you will create an Azure Sql Db and write output to a new table

Module 3: Manipulating Data

In this module you will learn how to manipulate data using the workflow tasks in MLStudio. This will include, cleaning data, joining datasets, adding columns, filtering using expressions and adding metadata to columns.

In this lab you will work with filters

In this lab you will orchestrate data flow with the forest fires dataset

In this lab you will build a count table to enable you to determine whether a movie is liked or not

Module 4: Statistical Analysis

In this module you will learn about math operations, linear correlation and hypothesis testing. This will enable you to whether or not the results of your experiments are statistically significant or not.

In this lab, you will investigate Linear Correlation.

In this lab, you will create investigate the user of averages in AzureML and construct a normal distribution function.

In this lab, you will investigate Linear Correlation.

Module 5: Testing and training data with Regression and Classification

In this module you will learn how to perform regression analysis to predict a continuous variable. You will also look at classifiers and how you can predict two or more discrete classes of output. You will learn how to analyse the results through a ROC curve and the coefficient of determination to enable a feedback loop in improving your model.

In this lab, you will use Automobile data such as the width and height of a car to predict its price.

In this lab, you will use Automobile data such as the width and height of a car to predict its price.

In this lab, you will publish the Regression model as a web service.

Module 6: Unsupervised learning with K-Means clustering

In this module you’ll learn how to use K-Means clustering to determine whether data points belong to a particular cluster. You’ll learn how to use clustering to build a pipeline of machine learning models which can be enriched to give better results together than separately.

In this lab, you will use K-Means clustering to take an unsupervised learning approach to seeing how Iris plants cluster together based on their measurements

Module 7: Collaborative Filtering with AzureML

In this module you’ll learn how to build a recommender system given a set of ratings and features for users and products. You’ll be able to choose which movies to watch given the genres you like and ratings and what others are watching and choose which things you should buy from a product catalogue given what others have bought.

In this lab, you will work out how to build a recommendations engine in AzureML.

In this lab, you will work out how to build a recommendations engine in AzureML.